Hi everyone, in terms of costs of hardware, would it be right to say that to Yugabyte is cost effective only when it comes to large scale usage compared to small to medium usages? If not can explain please?
Yes, for small dbs, a single node PostgreSQL will be more efficient. But you have to think in TCO (total cost of ownership). You might need other features that PostgreSQL doesn’t provide, like async-replication, multi-region-synchronous replication,etc.
While the YCQL (Cassandra) layer should be more efficient than Cassandra at all times.
It’s in the roadmap to have postgresql-performance parity even for small dbs.
Can you help me understand with an example?
Which case exactly? Some features:
- Storing rows clustered on the primary key
- Block compression on rows
- Tablet-design allows hosting vast amounts of data in a single node AND easier rebalancing (blog post)
- HASH & Range Sharding for tables & indexes